Classify strata

We assume there is a slight difference in the average value of the gene expression for a group of genes between two strata/time periods. There is no prior knowledge on which genes where the gene expression differs between the two strata. We test out the power of a large number of test statistics in order to identify whether the observed difference between the averages is significant. Secondly, we test out a large number of test statistics ability to classify new patients into the two strata based on the genes identified in the first analysis. This is performed both with data from one time-period and for cases where there is data from several time periods. Data from several time periods may be valuable if we expect a time development relative to time to diagnosis that is characteristic for the different strata.